基于集成算法的微博评论情感分析  

Sentiment Analysis of Weibo Comments Based on Intergrated Algorithm

作　　者：王德广[1] 黄丹婷 WANG Deguang;HUANG Danting(Software Technology Institute,Dalian Jiaotong University,Dalian 116028,China)

机构地区：[1]大连交通大学,软件学院,辽宁大连116028

出　　处：《微型电脑应用》2022年第10期134-137,共4页Microcomputer Applications

摘　　要：通过实验发现传统神经网络机器学习模型算法的准确率评估指标低于深度学习模型,但其在效率上却领先于深度学习模型,对此提出一种将传统机器学习模型与深度学习模型中各项指标最高的模型相结合的方法,使模型的准确率和效率都能得到有效的提升。选取新浪微博评论数据,提出将传统机器学习算法中的决策树、朴素贝叶斯和逻辑回归的集成模型与深度学习算法中RCCR模型即RNN、CNN和RCNN的集成模型相结合,实验结果表明与原算法相比,模型准确率达到87.55%,并且算法效率也得到了大幅提升,这证明了方法的有效性。Through experiment,it is found that the traditional neural network model of machine learning algorithm accuracy evaluation index is lower than the deep learning model,but its efficiency is better than the deep learning model.Thus,we put forward a model by combining the traditional machine learning and deep learning model in the indicators of the highest model,and the model accuracy and efficiency can be effectively promoted.Based on Sina Weibo review data,this paper proposes to combine the integration model of decision tree,naive Bayes and logistic regression in traditional machine learning algorithm with the integration model of RCCR model in deep learning algorithm,namely RNN,CNN and RCNN.The experimental results show that compared with the original algorithm,the accuracy rate of the model reaches 87.55%,and the efficiency of the algorithm has been greatly improved,which proves the effectiveness of the method.

关 键 词：机器学习 新冠肺炎 集成算法 情感分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]